▲ 243 points
back
70 comments
Was the eating of chocolate coins factored in? I think you can consider the reduction of chocolate coins due to consumption as theta . My initial assumption would be that there would be increased theta dependent on proximity to dinner.
Plus you could factor in gamma as the rate of change of the value of those coins as gummy bears are within proximity.
Just sayin' if you wanted to be real technical....
What quality of chocolate coins are we talking about? Are these Palmer quality or Lindt quality?
If they’re legally called “chocolate”, that’s a step up from Palmer.
Conversely, in an age before TV or Internet, the longer the game could keep kids occupied the better, no?
We might even imagine that anthropologically given the role of keeping kids occupied there were other games that were more fun but of a shorter duration which died out because parents instead pushed for teaching their kids the cultural equivalent of Monopoly.
Kids will typically play dreidel once in their life and then refuse to play it again "it's boring".
Instead the fun part is trick shots - spin it upside down, longest spin, start the spin high in the air and throw it still spinning, how many can you keep spinning simultaneously, spin a dreidel into the other player's dreidel. Things like that.
I dunno if I'd agree. If you asked a group of kids to play "see who can find the first prime greater than 2^201" then, per your criteria, that sounds great on paper but the kids will promptly discover the game sucks worse than massaging grandma's feet, quit playing, and never believe your game recommendations again. On the other hand, if a game is short but fun then it's likely it'll be played more than a single time per sitting on top of still be played in future sittings. The only key additional filter I can think of that is the game needs to not be so repetitive it is only fun the first time you play it.
After that it doesn't much matter if it takes 5 minutes and you play 10 times of 50 minutes and you play once - it just matters you had enough fun to entice you to keep playing instead of trying to find something else to do. How much fun that requires likely varies greatly by era though.
My head-cannon for Dreidel was that it invented by very bored people waiting out a siege during the Maccabean Revolt. That would explain the length of the game.
Apparently it’s not nearly that old.
Monopoly doesn’t take forever if you play by the rules.
https://bowilliams.com/2007/03/you-have-won-second-prize-in-...
My first thought for “can’t generalize over number of players” was that he could write code that emits a PRISM program…
Passing the state from n to n-1 players is harder.
This is also what I did when I used PRISM for my thesis. Honestly also not the greatest experience, but I think it does show that the limitations of the PRISM language are not inherent to the problem domain.
I love this, and have similar feelings about PRISM. I hope recent work to add probabilistic properties to TLA+ and P provide a route towards doing this kind of work with languages that are easier to use.
While this example is a bit silly, the ability to check properties like availability, latency, etc alongside the traditional liveness and safety properties is super useful for distributed systems work.
"Was this worth it?
Ugh no I hate PRISM"
:)
... that being said, as someone who only has a passing knowledge of the Dreidel the article held my interest
>So why bother with all this when the Python would be like ten lines?
Why not just write the Python to write this? Seems like you could add things like functions and generalizations to your meta-programming language and generate simple PRISM.
Home rules adjustment to make a game of Dreidel conclude in a reasonable amount of time: The ante increases by one each time the spin returns to the starting player. Getting a Shin and adding to the pot matches the current ante.
Adopted this from poker tournament's increasing blinds.
I'll likely start diving into probabilistic formal methods in a few months and this seems like a good place to ask: How does Storm [1] compare to PRISM?
It looks like a promising alternative and the academic pedigree seems top-notch, but I would like to know more from anyone with hands-on experience with it.
Hi lou, I'm a contributor to Storm and will continue to work on it in my upcoming PhD position. The bottom line is that there are some things you can only do in PRISM and there are more things you can only do in Storm, but if you can do it in both, Storm is usually much faster (see QComp: https://qcomp.org/).
PRISM has better documentation. Storm has recently recieved a grant that will help improve its user interface and applicabilty.
I would like to hear from you what ideas and applications you have for using PMC! :)
Finally, someone calls out the lie that is Big Dreidel.
A Wikipedia gem:
"Robert Feinerman has shown that the game of dreidel is unfair, in that the first player to spin has a better expected outcome than the second player, and the second better than the third, and so on"
Not entirely sure why such a simple game requires pulling out as big a gun as PRISM seems to be.
Looks to me like a very basic monte carlo simulation of Dreidel should take less than a 100 lines of C and would pretty much produce the exact same outcome.
Am I missing something? Does PRISM do formal, complicated things a brute force MC attack can't touch?
The article is literally called "Formally modeled Dreidel for no good reason" so that should answer your question.
From the beginning of the article:
> In 2015 Ben Blatt simulated the game and found that for P=4 and N=10, the average game will take 860 spins to finish.
> But this was based on a mere 50,000 simulated games! We can do better than a mere simulation. We can mathematically model the game and get exact numbers. So we're busting out PRISM.
While I was reading the post I had the opposite thought: isn't this some kind of Ehrenfest model? It shouldn't be too hard to formalize with pen and paper.
>Traditionally this game is played by kids, as it's an excellent way to convince them that gambling is boring as all hell.
I’m amazed it has endured this long.
New children are born every day. I'm not jewish, but I had jewish friends who taught me.
When I play dreidel you ante every turn… it does not take 860 spins.
To really teach people about gambling, try a variation where everyone someone antes, they ante a second coin to the "house".
Nice Read!
Looks like a fun game especially when using chocolate coins.
But I can see how that can get boring after a while and if finished there will be a lot of upset children. I think the winner will have to run for his life :)
PROTIP: make everyone antes
every single time, instead of just when the pot is empty. Quicker eliminations and more exciting pots.
Simulating a game you hate in a language you also hate, does that cancel out or double up?
If double negatives cancelled regress we'd drown in mediocrity. (Failure is a feature.) If a double positive cancelled progress we would stand next to the giants rather than on their shoulders.
This is why in our family there is an ante which increases every couple rounds...
Is there a way to make Dreidel more fun? Presumably some aspect of strategy, and perhaps a bit like lier's dice (or poker) in that there could be bluffing.
No Limit Texas Dreidel sounds like an improvement:
According to the author, game is designed to convince you that gambling isn't fun. Let's not corrupt that social good with technical fixes.
>Traditionally this game is played by kids, as it's an excellent way to convince them that gambling is boring as all hell.
More interesting IMNSHO than this post itself is the teaser for a future post:
"Next week will be the last newsletter of the year. Quick updates on this year's projects: ... Why programming languages doesn't have graph types: Done with interviews, writing an outline!"
I've always assumed that Dreidel is more like Poker than it is like War. In other words, it's a game designed around betting that is no fun when you take the betting out, rather than a game that is just a miserable time waster that only children enjoy.
Well written. I thoroughly enjoyed this. Is PRISM more or less the state of the art as far as software to analyze/model probabilistic systems?
I'd probably prefer to Markov chain the problem manually over that PRISM ugliness...
I live in Israel/Palestine, and strangely enough - the game is not played here AFAICT.
I mean, we have dreidels (they're called "svivonim" in Hebrew, from the root S, B/V, B/V , relating to rotation: "histovev" means "rotate" or "turn around", and "sivuv" is a rotation or a round) ; and we also have those chocolate coins - but they're not really related to each other. Frankly I have no idea how the chocolate coins relate to Hannukah at all, but whatever.
Anyway, the play with the dreidel is trying to get it to last as much time as you can without falling over, and that's basically it. There are fancy dreidels with colored lighting and a structure which increases the rotation time, that was a kind of a thing when I was younger at least. But the gambling is weird.
While writing a formal dreidel model might make you a better theorist, it won't make you a better dreidel spinner. In fact, it might make you worse at spinning dreidels, because it embodies the disastrous "formalism above all" doctrine.
Spinners like to obsess about gimmels and shins and other high-theory stuff, yet the following skills, essential for spinning dreidels, are almost never discussed and even more rarely practiced:
- Deciding how to grip the dreidel
- Proper vertical positioning so it doesn't immediately tumble
- Waiting until the clay is dry and ready, before you play
In that spirit, I offer the following alternative challenge: Create your own dreidel out of clay. Don't bother with formal theory, others have already done that for you. "Just" make a dreidel that can actually spin, even if it only supports a subset of the game at the beginning.
Is this a new hackernews meme?
It's the same format as another comment about writing compilers.
Nah I’ll just 3D print it. It’s almost 2024 baby!